Enhancing HIV Testing Indicator Reporting:

Data Science Approach for Identifying PMTCT Reporting Discrepancies

Abstract

In the landscape of healthcare data, consistency in reporting crucial indicators like HIV testing rates remains a challenge. Discrepancies in reporting, especially concerning Prevention of Mother-to-Child Transmission (PMTCT) tests, pose significant hurdles in understanding the true scope of HIV testing coverage. Our project aimed to address this issue using Data Science and AI methodologies. The Context: Within healthcare systems, the Comprehensive Patient Information Management System (CPIMS) is an essential tool for monitoring and evaluating health programs. However, inconsistencies in reporting HIV testing indicators, particularly concerning PMTCT sites, have hindered accurate assessments of testing coverage. The Problem: Facilities' varying inclusion/exclusion criteria for PMTCT tests and inconsistencies across testing locations have led to discrepancies in HIV testing indicator reporting. This hindered accurate estimation of HIV testing coverage and understanding of PMTCT site reporting behavior. Solution Approach: Utilizing Data Science and AI, our team developed a classification model. This model identifies PMTCT sites with reporting gaps, aiding in the identification of sites not reporting tests accurately. Leveraging machine learning algorithms, the model distinguishes different data behaviors, advising on estimates where data gaps exist and automating data cleanup processes. Outcome: The developed classification model offers a systematic approach to identify PMTCT sites that inaccurately report HIV testing data. This tool enables healthcare administrators to pinpoint specific facilities or locations with reporting discrepancies, facilitating targeted interventions and improving data accuracy. Next Steps: Further refinement of the model for real-time data integration, collaborating with healthcare facilities to implement automated data cleanup processes, and expanding the model's capabilities to encompass broader healthcare indicators.
Keywords: HIV Testing Indicator, PMTCT, Data Science, AI, Classification Model, Healthcare Data Accuracy, Data Behavior Identification.